Re: [GIT PULL] fuse fix for 2.6.39

[Linus Torvalds]

On Thu, May 12, 2011 at 10:46 AM, Trond Myklebust
<Trond.Myklebust@xxxxxxxxxx> wrote:
>
> I was thinking rather of just returning an EIO if people try. We know
> that there are breakages w.r.t. certain operations as I pointed out in
> my previous email.

Umm. Returning an error is ok, but only if you _need_ to return an error.

Returning an error because you don't like it - that's just wrong.

So I would seriously suggest you return an error only when you really
need something from the nd. So I'd suggest
 - for the automount case, return an error, since you can't change the filp
 - for the LOOKUP_RCU case, just do the "nd && (nd->flags & LOOKUP_RCU)"

The really doesn't seem to be many cases that really need the nd, so
why make it a hard error? *Most* users of nd already check for NULL,
either directly or indirectly (ie "if (!is_atomic_open(nd)) goto
no_open")

                   Linus
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/